Groq Free code examples in Python and JavaScript
Open models served on custom hardware — the fastest tokens per second available.
WHAT THIS PAGE COVERS
Streaming, where the speed difference is the whole point.
openai/gpt-oss-20b
groq
Model ids change. If a sample returns a 404 on the model, that is the first thing to check against the official docs.
Groq has its own SDK, and it is the OpenAI SDK's shape with a different name — same methods, same typed errors. You can also point the OpenAI client at Groq if you would rather not add a dependency.
pip install groq
export GROQ_API_KEY="gsk_..." # macOS / Linux
setx GROQ_API_KEY "gsk_..." # Windows, then open a new shellNote the model id has a namespace. Groq serves open models from several publishers, so ids look like openai/... or meta-llama/... — the prefix is part of the name, not a path.
import os
from groq import Groq
client = Groq(api_key=os.environ["GROQ_API_KEY"])
response = client.chat.completions.create(
model="openai/gpt-oss-20b",
messages=[
{"role": "system", "content": "You are a concise technical assistant."},
{"role": "user", "content": "Explain a database index in two sentences."},
],
)
print(response.choices[0].message.content)Groq's pitch is tokens per second, and non-streaming throws that away: the user waits for the full answer either way. Measure it, because the difference is the reason to be on this provider.
import time
start = time.perf_counter()
tokens = 0
stream = client.chat.completions.create(
model="openai/gpt-oss-20b",
messages=[{"role": "user", "content": "Explain B-trees in 200 words."}],
stream=True,
)
for chunk in stream:
delta = chunk.choices[0].delta.content
if delta:
tokens += 1
print(delta, end="", flush=True)
elapsed = time.perf_counter() - start
print(f"\n~{tokens / elapsed:.0f} chunks/sec")Because the SDK mirrors OpenAI's, the same code runs against either — which is what makes Groq easy to trial and easy to fall back from.
import os
from groq import Groq
from openai import OpenAI
def build_client():
# One switch. Everything downstream calls .chat.completions.create
# with the same arguments on both clients.
if os.environ.get("USE_GROQ"):
return Groq(api_key=os.environ["GROQ_API_KEY"]), "openai/gpt-oss-20b"
return OpenAI(), "gpt-5.4-mini"
client, model = build_client()
response = client.chat.completions.create(
model=model,
messages=[{"role": "user", "content": "Explain a database index."}],
)
print(response.choices[0].message.content)
# The typed errors line up too: groq.RateLimitError and
# openai.RateLimitError have the same name and the same meaning.Groq's own SDK mirrors the OpenAI SDK's shape. Server-side only.
npm install groq-sdk
export GROQ_API_KEY="gsk_..."Model ids carry a publisher prefix — openai/... or meta-llama/... — because Groq serves open models from several sources. The prefix is part of the id.
import Groq from "groq-sdk";
const client = new Groq({ apiKey: process.env.GROQ_API_KEY });
const response = await client.chat.completions.create({
model: "openai/gpt-oss-20b",
messages: [
{ role: "system", content: "You are a concise technical assistant." },
{ role: "user", content: "Explain a database index in two sentences." },
],
});
console.log(response.choices[0].message.content);Non-streaming discards the reason to be here: the user waits for the whole answer regardless of how fast it was generated.
const started = performance.now();
let chunks = 0;
const stream = await client.chat.completions.create({
model: "openai/gpt-oss-20b",
messages: [{ role: "user", content: "Explain B-trees in 200 words." }],
stream: true,
});
for await (const chunk of stream) {
const delta = chunk.choices[0]?.delta?.content;
if (delta) {
chunks++;
process.stdout.write(delta);
}
}
const seconds = (performance.now() - started) / 1000;
console.log(`\n~${Math.round(chunks / seconds)} chunks/sec`);The two SDKs share a surface, so the switch is where you build the client and nowhere else.
import Groq from "groq-sdk";
import OpenAI from "openai";
function buildClient() {
if (process.env.USE_GROQ) {
return [new Groq({ apiKey: process.env.GROQ_API_KEY }), "openai/gpt-oss-20b"];
}
return [new OpenAI(), "gpt-5.4-mini"];
}
const [client, model] = buildClient();
const response = await client.chat.completions.create({
model,
messages: [{ role: "user", content: "Explain a database index." }],
});
console.log(response.choices[0].message.content);QUESTIONS
Which languages are the Groq examples in?
Python and JavaScript. Every sample runs as pasted and reads its credentials from an environment variable — no key is ever written into the code.
Which package do I install for Groq?
The samples use groq. The install step on each language tab gives the exact command.
Which Groq model do these examples use?
openai/gpt-oss-20b, checked against the official documentation on 25 September 2026. Model ids change often; if a call returns a model-not-found error, check the current list in the official docs.
Where is the official Groq documentation?
https://console.groq.com/docs — it is linked at the bottom of this page, and it is the authority whenever it disagrees with a sample here.
ALSO IN LANGUAGE MODELS
The authority
Official Groq documentation
Model names, pricing and rate limits change faster than any page here can. When this page and the official docs disagree, the official docs are right.
https://console.groq.com/docs